Feature fusion method based on multi-kernel learning

Abstract

The invention belongs to the technical field of multimode composite guidance information fusion, and discloses a feature fusion method based on multi-kernel learning, which specifically comprises thefollowing steps: 1) performing standardization processing on radar and infrared training feature sets respectively; 2) respectively extracting nonlinear features X (f) and Y (g) of radar and infraredby using neural networks f and g; 3) constructing a fusion criterion function on an output layer of the neural network to maximize a correlation coefficient corr (X (f), Y (g)); 4) optimizing the neural network to obtain a fused feature vector; 5) determining a pre-selected basis kernel function; 6) obtaining a synthetic kernel by adopting a weighted summation mode; 7) training a synthetic kernelby using a simple multi-kernel learning algorithm; 8) repeating the steps 1-4 for the radar and infrared test feature set during online identification, and replacing a single kernel function in a traditional support vector machine with the trained synthetic kernel; and 9) confirming the identity of the to-be-attacked target, and reducing the feature dimension and improving the identification performance of information fusion while ensuring the maximum inter-class scatter matrix and the minimum intra-class scatter matrix.

Description

technical field [0001] The invention relates to the technical field of multi-mode composite guidance information fusion, in particular to a feature fusion method based on multi-core learning, which can be used to confirm the identity of non-cooperative targets in radar and infrared composite guidance. Background technique [0002] The multi-mode compound homing technology has become a mainstream research direction of precision guided weapons. The millimeter-wave radar seeker has a long detection range and can work around the clock, but it is vulnerable to electronic interference and electronic deception. Although the infrared thermal imaging seeker has high detection accuracy and strong anti-interference ability, its detection distance is short. The millimeter-wave radar / infrared thermal imaging dual-mode composite guidance system uses the advantages of each single mode to combine detection, which can learn from each other's strengths and make up for the defects and shortcom...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problems existing in the prior art, the purpose of the present invention is to provide a feature fusion method based on multi-core learning, which solves the problem that in the training phase of the existing feature fusion algorithm, all feature components are mapped with the same kernel function, so that the feature In order to solve the problem of limited separability, under the premise of considering the influence of kernel function selection on the performance of feature fusion algorithm, and using the radar / infrared composite seeker to identify the ground target as the application background, the feature layer fusion method is compared with the kernel layer fusion method. In combination, a feature fusion method based on multi-core learning is proposed on the basis of the deep canonical correlation analysis algorithm, which improves the target identification performance of the information fusion system and can be used to confirm the identity of non-cooperative targets in radar and infrared composite guidance; While the inter-class scatter matrix is ​​the largest and the intra-class scatter matrix is ​​the smallest, the feature dimension can be reduced, which is conducive to further improving the identification performance of the information fusion system

Images